IMPROVED MODULATION TECHNIQUES FOR

MATRIX CONVERTERS AND THEIR APPLICATIONS IN

INDUCTION MOTOR DRIVES

TABISH NAZIR MIR

DEPARTMENT OF ELECTRICAL ENGINEERING

INDIAN INSTITUTE OF TECHNOLOGY DELHI
NOVEMBER 2020
© Indian Institute of Technology Delhi (IITD), New Delhi, 2020
 IMPROVED MODULATION TECHNIQUES FOR
MATRIX CONVERTERS AND THEIR APPLICATIONS IN
INDUCTION MOTOR DRIVES

by

TABISH NAZIR MIR

Department of Electrical Engineering

Submitted
in fulfillment of the requirements of the degree of Doctor of Philosophy
to the

INDIAN INSTITUTE OF TECHNOLOGY DELHI

NOVEMBER 2020
 CERTIFICATE

This is to certify that the thesis entitled ‘Improved Modulation Techniques for
Matrix Converters and their Applications in Induction Motor Drives’ being
submitted by Ms. Tabish Nazir Mir for award of the degree of Doctor of Philosophy
in the Department of Electrical Engineering, Indian Institute of Technology Delhi, is a
record of the student work carried out by her under our supervision. The matter embodied
in this thesis has not been submitted for award of any other degree or diploma.

Dated: November, 2020

(Prof. Bhim Singh)

Supervisor
Electrical Engineering Department
Indian Institute of Technology Delhi
Hauz Khas, New Delhi-110016, India

(Prof. A. H Bhat)
Co-Supervisor
Electrical Engineering Department
National Institute of Technology Srinagar
Hazratbal, J&K-190006, India

i
 ACKNOWLEDGEMENTS

I would like to express my earnest gratitude to my supervisor, Prof. Bhim Singh and
co-supervisor, Prof. A.H Bhat, for their continuous guidance, support and motivation.
Prof. Bhim Singh has not only been an inspiration, but a guiding light and a visionary
mentor. His ingenuity and enthusiasm for research, has been a driving force in keeping me
motivated throughout my doctoral studies, and he continues to inspire my career ahead.
It has been an exhilarating experience to work under his supervision, and to witness his
commitment and passion towards his work. Apart from sharing technical nuances and a
multitude of ideas, Prof. Bhim Singh has taught me the importance of discipline, tenacity
and perseverance. It has been an absolute honor to learn from him, to be part of his
research group, and to do my bit in carrying forward the legacy that he has created. For
his teachings, and for his faith in me, I am forever indebted.
The daunting task of pursuing Ph.D. from IIT Delhi and simultaneously serving as
faculty at NIT Srinagar, would not have been accomplished without the unflinching sup-
port of my supervisor, Prof. A.H Bhat at NIT Srinagar. From introducing me to power
electronics in my undergraduate studies, to motivating and supporting me throughout
my Ph.D. journey, Prof. A.H Bhat has been instrumental in the completion of this the-
sis. Without the lab infrastructure created by Prof. A.H Bhat at NIT Srinagar, this
thesis in its current form would not have been possible. In my moments of despair, it
has been Prof. Bhat who has reminded me that there is no substitute for hardwork and
that my hardwork shall pay off. For those reminders, and for his unwavering support and
guidance, I am forever indebted.
I wish to express my sincere thanks and heartfelt gratitude to all my SRC mem-
bers, Prof. Sukumar Mishra, Dr. Anandarup Das, and Dr. Ashu Verma, for
their valuable inputs, regular guidance and constructive suggestions that have helped in
shaping this thesis.
I sincerely thank the Department of Electrical Engineering at NIT Srinagar, for lend-
ing financial and laboratory support to my research, especially Prof. S.A Lone and Prof.
A.H Bhat, who extended complete financial support during their respective tenures as

ii
Heads of the department. I am eternally grateful to Prof. Bhim Singh, for his financial
support towards experimentation at IIT Delhi, conference registrations and article pro-
cessing charges. My sincere thanks to the lab staff, Mr. Puran Singh, and Mr. Jitender
Singh for facilitating my work at PG Machines Lab, IIT Delhi, and Mr. Farooq for his
assistance during experimentation at Power Electronics Lab, NIT Srinagar.
Special thanks to my lab mates at IIT Delhi, and colleagues at NIT Srinagar. It has
been a pleasure to have the company of Mrs. Subarni Pradhan, Mrs. Nidhi Mishra, Mr.
Anshul Varshney, Mr. Utkarsh Sharma, Mrs. Farheen, Mrs. Shubra, Mr. Piyush Kanth,
Mrs. Pavitra, Mrs. Rohini, Ms. Hina Parveen, Mrs. Rashmi, Dr. Nishant Kumar, Dr.
Aniket Anand, Ms. Yashi, and all juniors and seniors at P.G Machines Lab. A special
mention to Mrs. Subarni for being my constant companion and friend at IIT Delhi, and
to Mr. Anshul Varshney for being one of the most helpful people in the lab. My deepest
gratitude to my friends and colleagues at NIT Srinagar, Dr. Fatima Jalid, Ms. Aravi
Muzzaffar, Mrs. Uferah Maqbool, Ms. Iqra, Ms. Falak, Mr. Ved Prakash and Dr. Janib
Bashir, who have been partners with me in handling full time teaching while pursuing
Ph.D.
So often you find that the students you’re trying to inspire, end up inspiring you.
A very special mention goes to my dearest students who continue to make me proud,
especially those who did their final year B.Tech projects with me, and with whom I
shared my struggles, failures and victories at experimental work. My deepest gratitude
to Sohaib, Adnan, and Burooj, for sharing my excitement and zeal for matrix converters,
and for helping me in experimentation.
I express my profound gratitude to my parents for their prayers and encouragement.
If it wasn’t for my father, Mr. Nazir Ahmed Mir, I would not have known sincerity, and
it is my mother, Mrs. Masooda Nazir, who taught me endurance. These two virtues have
been my guiding stars during this journey. I am eternally grateful to my parents for their
incredible support system and my sisters Dr. Mahvish Nazir and Dr. Beenish Nazir for
being my biggest cheer leaders.
My deepest appreciation is for my husband, Mr. Aamir Rafiq, for his unfaltering

iii
support towards my dreams and aspirations. His immeasurable patience, love and reas-
surance, has gone a long way in helping me overcome the difficulties that I encountered
during the course of my research. Thank you for always having my back, and for making
my dreams, your own.
Finally, my most sincere gratitude to the Almighty, who has known all my moments
of joy and despair, and has given me the strength to never say die.

Date: November 2020

Tabish Nazir Mir

iv
 ABSTRACT

One of the major industrial applications of power electronics, is the control of motor
drives, which are mostly dominated by converter fed induction motors. In such appli-
cations, features like energy efficiency, improved power quality, high power density and
improved reliability are highly desirable. While most three-phase induction motor drives
are conventionally fed through AC-DC-AC converters, this traditional approach is as-
sociated with a large footprint owing to an intermediate DC stage. DC link capacitor
used as an intervening energy storage element, is not only bulky but also contributes to
low system reliability. An interesting alternative in the form of matrix converters, has
been widely researched over the past two decades. Matrix converter belongs to the breed
of direct AC-AC converters, that facilitate multi-phase power conversion with smooth
voltage control at unrestricted frequency, without any intervening DC link. It offers
advantages such as improved power quality on source as well as load side, inherent bi-
directionality, completely controllable input power factor, compact size and possibility
for common mode voltage elimination. Owing to these features, matrix converters have
been used in motor drives, wind energy conversion systems and aerospace applications.
In spite of a wide range of advantages, the industry acceptability of matrix converters
has been limited. This is primarily because of difficulties in control, protection, and
commutation. While some of these challenges have been overcome through research,
others are still being worked upon. This thesis deals with the design and development
of matrix converter fed induction motor drive, with complete protection and multi-step
commutation. It elucidates some novel and improved modulation strategies for matrix
converters, from the perspective of controlling induction motor drives. For the purpose of
controlling three-phase induction motor drive, two topologies are selected viz. three-phase
to three-phase matrix converter and single-phase to three-phase matrix converter. The
challenges associated with the modulation of each topology are individually addressed,
and their suitability is explored in the speed sensor-less control of three-phase induction
motor drives.
Three-phase to three-phase matrix converter is the most commonly explored topol-

v
ogy amongst multi-phase matrix converters. A large number of possible switching states,
coupled with the simultaneous control of source and load side parameters, leads to com-
plicated modulation. Space vector modulation remains one of the most popular control
techniques which allows concurrent tracing of output voltage and input current space
vectors. However, it is associated with high processor burden and complex implementa-
tion. Over the past decade, finite control set model predictive control (FCS-MPC) has
been explored in the modulation of three-phase to three-phase matrix converter. It has
especially shown promising results while achieving multiple objectives of source and load
current control. However, it is associated with extensive mathematical modeling of load
and input filter, besides performance dependence on processor strength. This thesis deals
with certain improved and novel modulation approaches for three-phase to three-phase
matrix converter that tend to simplify modulation and control. It deals with an im-
provised multi-objective model predictive control strategy that combines FCS-MPC and
space vector modulation to achieve enhanced system performance at higher sample times
and limits switching frequency variations. Furthermore, a unique predictive delta sigma
modulation (PDSM) technique is proposed, that is simple and intuitive to implement,
with significantly reduced processor burden and enhanced system performance as com-
pared to previously documented techniques. PDSM is also explored in common mode
voltage elimination for three-phase to three-phase matrix converter based systems. Speed
sensor-less vector control is implemented for commonly used 21 states of the converter
using PDSM technique with a full order predictive observer for wide range of speed es-
timation. Moreover, the matrix converter fed induction motor drive is modulated using
only six common mode voltage eliminating states in order to present a more reliable drive
solution, while tapping all other advantages of the converter, over a wide range of rotor
speed. The speed sensor-less drive is also tested for hysteresis based direct torque control
(DTC) and constant switching frequency based DTC, with suitable choice of speed ob-
servers. All control algorithms are developed in Matlab/Simulink environment, followed
by experimental verification on a four layer PCB prototype of matrix converter based
induction motor drive.

vi
 In order to facilitate the control of a three-phase induction motor on a single phase sup-
ply, the use of single-phase to three-phase matrix converter is advocated. This topology
has not garnered significant interest of the research community owing to a poor voltage
conversion ratio, and severely compromised harmonic performance due to instantaneous
power mismatch between single-phase input and three-phase output. This thesis explores
the optimum use of single-phase to three-phase matrix converter for low speed sensor-less
control of three-phase induction motor drive from a single-phase supply. The delta sigma
technique is demonstrated for enhancement of motor current quality in the low speed
region, while compromising on the quality of source current. Further, a multi-objective
FCS-MPC technique is demonstrated and compared with other modulation schemes, to
achieve a suitable trade-off between the power quality on the grid side versus that on
the motor side. All modulation techniques are experimentally validated on single-phase
to three-phase matrix converter fed inductive load followed by demonstration with speed
sensor-less control on a three-phase induction motor drive.

vii
viii
ix
Table of Contents

List of Figures xviii

List of Tables xxx

CHAPTER I INTRODUCTION 1

1.1 General 1

1.2 State of Art on Static AC-AC Power Conversion 2

1.2.1 AC-DC-AC Converters 2
1.2.2 Cyclo-Converters 3
1.2.3 Matrix Converters 4

1.3 Objectives and Scope of Work 7

1.4 Outline of Chapters 9

CHAPTER II LITERATURE REVIEW 13

2.1 General 13

2.2 Literature Survey 13

2.2.1 Review of Matrix Converter Topologies 14
2.2.1.1 Three Phase to Three Phase Matrix Converters 14
2.2.1.2 Single Phase to Three Phase Matrix Converters 16
2.2.2 Review of Modulation Strategies for Matrix Converters 18
2.2.2.1 Venturini-Alesini Algorithm for Matrix Converters 19
2.2.2.2 Scalar Control Algorithm for Matrix Converters 20
2.2.2.3 Space Vector Modulation for Matrix Converters 20
2.2.2.4 Model Predictive Control for Matrix Converters 21
2.2.3 Review of Protection and Commutation Techniques for Bi-Directional
Devices Used in Matrix Converters 22
2.2.4 Review of Matrix Converter Fed Induction Motor Drives 24
2.2.5 Review of Sensor-less Control of Induction Motor Drives 25

2.3 Identified Research Areas 26

2.4 Conclusions 27

CHAPTER III CLASSIFICATION, CONFIGURATIONS AND DE-

SIGN OF MATRIX CONVERTERS 28

3.1 General 28

3.2 Classification of Matrix Converter Topologies 28

3.3 Configurations of Matrix Converter Topologies 29

x
 3.3.1 Three Phase to Three Phase Matrix Converter Fed Induction Motor
Drive 29
3.3.2 Single-Phase to Three-Phase Matrix Converter Fed Induction Mo-
tor Drive 30

3.4 Design and Implementation 30

3.4.1 Design and Layout of Power Circuit 31
3.4.2 Design and Layout of Gate Driver Circuits 32
3.4.3 Design and Implementation of Input Filter 32
3.4.4 System Auxiliaries 34
3.4.4.1 Signal Conditioning Circuit for Voltage Sensors 34
3.4.4.2 Signal Conditioning Circuit for Current Sensors 35

3.5 Selection of Induction Motor 35

3.6 Protection Issues in Matrix Converter 36

3.6.1 FPGA implementation of Four Step Commutation 37
3.6.2 Design of Clamp Circuit 38

3.7 Execution of Control Algorithm using dSPACE RTI 1202 39

3.8 Integrated Hardware Setup 40

3.9 Results and Discussion 41

3.9.1 Test Results of Four Step Commutation 41
3.9.2 Test Results of Voltage and Current Sensing Units 43
3.9.3 Test Results of Space Vector Modulation for Matrix Converter 43

3.10 Conclusions 44

CHAPTER IV IMPROVED MODULATION TECHNIQUES FOR MA-

TRIX CONVERTERS 45

4.1 General 45

4.2 Configuration of Three-Phase to Three-Phase Matrix Con-

verter Fed Inductive Load 46

4.3 Requirements of Modulation Techniques for Matrix Converters 47

4.4 Modulation Techniques for Matrix Converters 48
4.4.1 Multi-objective Finite Control Set Model Predictive Control (FCS-
MPC) for Matrix Converters 49
4.4.1.1 Classical Multi-objective FCS-MPC for Matrix Converter 49
4.4.1.1.1 Mathematical Modeling of Load 50
4.4.1.1.2 Mathematical Modeling of Input Filter 50
4.4.1.1.3 Weight Selection and Minimization of Cost Func-
tion 52
4.4.1.1.4 Limitations of Classical Multi-objective FCS-MPC 52
4.4.1.2 Fuzzy Decision Making (FDM) for Elimination of Weigh-
ing Factors 53

xi
 4.4.1.3 Improved Switching Implementation by Incorporating Space
Vector Modulation (SVM) in Fuzzy Decision Making (FDM) 54
4.4.2 Predictive Delta Sigma Modulation for Matrix Converters 56
4.4.2.1 Load Voltage Control 58
4.4.2.2 Source Current Control 59
4.4.2.2.1 Generation of Source Current Reference 59
4.4.2.2.2 Load Current Prediction for Source Current Con-
trol 60
4.4.2.3 Development of the Final Switching Scheme 62
4.4.2.4 Design of Integrator Gains, kv and ki 63
4.4.2.5 Effect of Hysteresis Band 64
4.4.3 Predictive Delta Sigma Modulation for Common Mode Voltage
(CMV) Elimination in Matrix Converters 65
4.4.3.1 Load Voltage Control for CMV Elimination 67
4.4.3.2 Source Current Control for CMV Elimination 68
4.4.3.3 Final Switching 69

4.5 MATLAB Based Modeling and Simulation of Three-Phase to

Three-Phase Matrix Converter Fed Inductive Load 71

4.6 Hardware Implementation of Three-Phase to Three-Phase Ma-

trix Converter Fed Inductive Load 72

4.7 Results and Discussion 72

4.7.1 Performance of Improved Switching Implementation for Matrix Con-
verter by incorporating Space Vector Modulation (SVM) in Fuzzy
Decision Making (FDM) 73
4.7.1.1 Load Side Performance 74
4.7.1.2 Source Side Performance 75
4.7.1.3 Power Quality Performance 76
4.7.2 Performance of Predictive Delta Sigma Modulation for Matrix Con-
verter 79
4.7.2.1 Dynamic Performance 80
4.7.2.2 Steady State Performance 82
4.7.2.3 Power Quality Performance 83
4.7.3 Performance of Predictive Delta Sigma Modulation for CMV Elim-
ination in Matrix Converter 84
4.7.3.1 Dynamic Performance 85
4.7.3.2 Steady State Performance 86
4.7.3.3 Power Quality Performance 88
4.7.4 Comparative Analysis of Modulation Techniques 90

4.8 Conclusions 92

CHAPTER V SPEED SENSOR-LESS VECTOR CONTROL OF THREE-

PHASE TO THREE-PHASE MATRIX CONVERTER FED INDUC-
TION MOTOR DRIVE 95

5.1 General 95

xii
5.2 Configuration of Three-Phase to Three-Phase Matrix Con-
verter Fed Induction Motor Drive 95

5.3 Vector Control of Three-Phase to Three-Phase Matrix Con-

verter Fed Induction Motor Drive 96
5.3.1 Speed Sensor-less Vector Control of Three-Phase to Three-Phase
Matrix Converter Fed Induction Motor Drive Using Conventional
21 States 97
5.3.1.1 State Estimation Using Model Predictive Control 98
5.3.1.2 Control and Modulation 102
5.3.1.2.1 Flux and Speed Control 102
5.3.1.2.2 Input-Output Power Balance 103
5.3.1.2.3 Delta Sigma Modulation 104
5.3.2 Speed Sensor-less Vector Control of Three-Phase to Three-Phase
Matrix Converter Fed Induction Motor Drive Using 6 Common
Mode Voltage Eliminating States 105
5.3.2.1 Multi-Objective FCS-MPC of Matrix Converter Fed In-
duction Motor With Zero CMV 109
5.3.2.1.1 Mathematical Modeling of Induction Motor for
Stator Current Prediction 109
5.3.2.1.2 Modeling of Input Filter for Source Current Pre-
diction 111
5.3.2.1.3 State Selection Using Fuzzy Normalization 112
5.3.2.2 State Estimation and Control 114
5.3.2.2.1 Estimation of Rotor Flux and Rotor Speed 114
5.3.2.2.2 Generation of Stator and Source Current Refer-
ences 116

5.4 MATLAB Based Modeling and Simulation for Speed Sensor-

less Vector Control of Three-Phase to Three-Phase Matrix
Converter Fed Induction Motor Drive 117

5.5 Hardware Implementation for Speed Sensor-less Vector Con-

trol of Three-Phase to Three-Phase Matrix Converter Fed In-
duction Motor Drive 118

5.6 Results and Discussion 119

5.6.1 Performance of Speed Sensor-less Vector Control for Three-Phase
to Three-Phase Matrix Converter Fed Induction Motor Drive Using
Conventional 21 States 119
5.6.1.1 Starting Performance 120
5.6.1.2 Dynamics During Speed Control and Reversal 120
5.6.1.3 Dynamics During Load Perturbation 122
5.6.1.4 Steady state Performance 122
5.6.1.5 Power Quality Performance 124
5.6.2 Performance of Speed Sensor-less Vector Control for Three-Phase
to Three-Phase Matrix Converter Fed Induction Motor Drive with
Zero CMV 124
5.6.2.1 Starting Performance 125

xiii
 5.6.2.2 Dynamics During Speed Control and Reversal 126
5.6.2.3 Dynamics During Load Perturbation 128
5.6.2.4 Steady State Performance 129
5.6.2.5 Power Quality Performance 129
5.6.2.6 Performance in terms of Common Mode Voltage 130
5.6.2.7 Comparative Assessment 132
5.6.3 Design of Control Gains 133

5.7 Conclusions 135

CHAPTER VI SPEED SENSOR-LESS DIRECT TORQUE CONTROL

OF THREE-PHASE TO THREE-PHASE MATRIX CONVERTER FED
INDUCTION MOTOR DRIVE 137

6.1 General 137

6.2 Configuration of Three-Phase to Three-Phase Matrix Con-

verter Fed Induction Motor Drive 138

6.3 Direct Torque Control of Three-Phase to Three-Phase Matrix

Converter Fed Induction Motor Drive 138
6.3.1 Speed Sensor-less DTC of Three-Phase to Three-Phase Matrix Con-
verter Fed Induction Motor Drive 139
6.3.1.1 DTC Through Matrix Converter 140
6.3.1.2 Input Power Factor Control 142
6.3.1.3 Speed Sensor-less Operation Through Adaptive Flux Ob-
server 144
6.3.1.3.1 Modeling of Induction Motor 145
6.3.1.3.2 Adaptive Flux Observer 146
6.3.2 Constant Switching Frequency DTC for Three-Phase to Three-
Phase Matrix Converter Fed Speed Sensor-less Induction Motor
Drive 147
6.3.2.1 State Estimation 149
6.3.2.1.1 Stator Voltage, Flux and Torque Estimation 149
6.3.2.1.2 MRAS based Speed Estimation 150
6.3.2.2 DTC at Constant Switching Frequency 151
6.3.2.3 SVM of MC with Input Power Factor Correction 154

6.4 MATLAB Based Modeling and Simulation for Speed Sensor-

less DTC of Three-Phase to Three-Phase Matrix Converter
Fed Induction Motor Drive 156

6.5 Hardware Implementation for Speed Sensor-less DTC of Three-

Phase to Three-Phase Matrix Converter Fed Induction Motor
Drive 156

6.6 Results and Discussion 158

6.6.1 Performance of Speed Sensor-less DTC for Three-Phase to Three-
Phase Matrix Converter Fed Induction Motor Drive 158
6.6.1.1 Starting Performance 158

xiv
 6.6.1.2 Dynamics During Speed Reversal 159
6.6.1.3 Dynamics During Load/Source Perturbation 160
6.6.1.4 Steady State Performance 161
6.6.1.5 Power Quality Performance 163
6.6.2 Performance of Constant Switching Frequency DTC for Three-
Phase to Three-Phase Matrix Converter Fed Induction Motor Drive
with Zero CMV 165
6.6.2.1 Starting Performance 165
6.6.2.2 Dynamics During Speed Reversal 167
6.6.2.3 Steady State Performance 167
6.6.3 Comparative Analysis of Constant Switching DTC with Conven-
tional DTC 169

6.7 Conclusions 170

CHAPTER VII LOAD ORIENTED CONTROL OF SINGLE-PHASE

TO THREE-PHASE MATRIX CONVERTER FED INDUCTION MO-
TOR DRIVE 172

7.1 General 172

7.2 Configuration of Single-Phase to Three-Phase Matrix Con-

verter Fed Induction Motor Drive 173

7.3 Low Speed Sensor-less Load Oriented Control of Single-Phase

to Three-Phase Matrix Converter Fed Induction Motor Drive 174
7.3.1 Delta Sigma Modulation for Load Side Power Quality Improvement
in Single-Phase to Three-Phase Matrix Converter 174
7.3.2 Full Order Predictive Observer for State Estimation 177
7.3.3 Speed Sensor-less Control 178

7.4 MATLAB Based Modeling and Simulation for Speed Sensor-

less Vector Control of Single-Phase to Three-Phase Matrix
Converter Fed Induction Motor Drive 178

7.5 Hardware Implementation for Speed Sensor-less Vector Con-

trol of Single-Phase to Three-Phase Matrix Converter Fed In-
duction Motor Drive 179

7.6 Results and Discussion 180

7.6.1 Performance of Delta Sigma Modulation for Single-Phase to Three-
Phase Matrix Converter 181
7.6.1.1 Dynamic Performance 181
7.6.1.2 Steady State Performance 182
7.6.1.3 Power Quality Performance 184
7.6.2 Performance of Speed Sensor-less Vector Control for Single-Phase
to Three-Phase Matrix Converter Fed Induction Motor Drive 185
7.6.2.1 Starting Performance 186
7.6.2.2 Dynamics During Speed Control and Reversal 187
7.6.2.3 Dynamics During Load Perturbation 187

xv
 7.6.2.4 Steady State Performance 190
7.6.2.5 Power Quality Performance 190

7.7 Conclusions 192

CHAPTER VIII MULTI-OBJECTIVE PREDICTIVE CONTROL FOR

SINGLE-PHASE TO THREE-PHASE MATRIX CONVERTER 194

8.1 General 194

8.2 Configuration of Single-Phase to Three-Phase Matrix Con-

verter Fed Induction Motor Drive 196

8.3 Modulation of 1-φ TO 3-φ Matrix Converter 197

8.4 Finite Control Set Model Predictive Control for 1-φ TO 3-φ
Matrix Converter 199
8.4.1 Load Oriented FCS-MPC for 1-φ TO 3-φ Matrix Converter 200
8.4.2 Compromise Between Load and Source Current Quality Using Multi-
Objective FCS-MPC 201

8.5 Speed Sensor-less Multi-Objective FCS-MPC for 1-φ TO 3-φ

Matrix Converter fed induction motor drive 202

8.6 MATLAB Based Modeling and Simulation for Speed Sensor-

less Vector Control of Single-Phase to Three-Phase Matrix
Converter Fed Induction Motor Drive 203

8.7 Hardware Implementation for Speed Sensor-less Vector Con-

trol of Single-Phase to Three-Phase Matrix Converter Fed In-
duction Motor Drive 204

8.8 Results and Discussion 204

8.8.1 Comparative Analysis of Modulation Techniques for Single-Phase
to Three-Phase Matrix Converter 205
8.8.1.1 Dynamic Performance 205
8.8.1.2 Steady State Performance 207
8.8.1.3 Power Quality Performance 209
8.8.2 Performance of MO-FCS-MPC Based Single-Phase to Three-Phase
Matrix Converter Fed Speed Sensor-less Induction Motor Drive 210
8.8.2.1 Starting Performance 211
8.8.2.2 Dynamic Performance 211
8.8.2.3 Steady State Performance 212
8.8.2.4 Power Quality Performance 212

8.9 Conclusions 215

xvi
CHAPTER IX MAIN CONCLUSIONS AND SUGGESTIONS FOR
FURTHER WORK 218

9.1 General 218

9.2 Main Conclusions 219

9.3 Suggestions for Further Work 221

REFERENCES 223

LIST OF PUBLICATIONS 240

APPENDIX 242

BIODATA 245

xvii
List of Figures

Figure 1.1 Conventional two stage AC-DC-AC converter fed induction mo-
tor drive 3
Figure 1.2 Single stage three-phase to three-phase half wave cycloconverter
fed induction motor drive 4
Figure 1.3 Three-phase to three-phase matrix converter fed induction motor
drive 5
Figure 1.4 Diagrammatic overview of work 9

Figure 2.1 Different configurations of bi-directional devices; (a) Common-

Collector configuration (b) Common-emitter configuration (c) Four diodes
and one IGBT in a bridge (d) Reverse blocking IGBTs in antiparellel. 23

Figure 3.1 Circuit configuration of 3-φ to 3-φ matrix converter fed induction
motor drive 30
Figure 3.2 Circuit configuration of 1-φ to 3-φ matrix converter fed induction
motor drive 31
Figure 3.3 Four layer PCB design of matrix converter prototype; (a) PCB
Layer 1 (b) PCB Layer 2 (c) PCB Layer 3 (d) PCB Layer 4. 32
Figure 3.4 Four layer PCB design of gate driver circuits; (a) PCB Layer 1
(b) PCB Layer 2 (c) PCB Layer 3 (d) PCB Layer 4. 33
Figure 3.5 Signal conditioning circuit for voltage sensor LEM LV 25P 35
Figure 3.6 Signal conditioning circuit for current sensor LEM LA55P 36
Figure 3.7 Illustration of four step commutation technique for positive current 38
Figure 3.8 Matrix converter enabled with over-voltage protection through
clamp circuit 40
Figure 3.9 Four layer PCB design of clamp circuit; (a) PCB Layer 1 (b)
PCB Layer 2 (c) PCB Layer 3 (d) PCB Layer 4. 41

xviii
Figure 3.10 Picture of the complete experimental setup 42
Figure 3.11 Experimental verification of four step commutation; (a) For de-
vices carrying positive current (b) For devices carrying negative current. 42
Figure 3.12 Test results for sensing units; (a) Test result for voltage trans-
ducer (b) Test result for current transducer. 43
Figure 3.13 Test results for space vector modulation; (a) Load side response
(b) Source side response. 44

Figure 4.1 3φ-3φ (ABC-abc) matrix converter fed inductive (RL) load 46
∗
Figure 4.2 Reference output voltage (vabc ) and reference input current (i∗ABC )
space vector diagrams in a 3φ-3φ MC 55
Figure 4.3 Complete block diagram for the implementation of improvised
Mo-FCS-MPC using fuzzy logic and sector information from space vectors 56
Figure 4.4 Block diagram representation of the complete modulation algo-
rithm 63
Figure 4.5 Hysteresis control of integrator outputs 63
Figure 4.6 Signal chasing for; (a) H = ±0.01 (b) H = ±0.1 65
Figure 4.7 Minimum switching transitions per state change; (a) Conven-
tional Group-I and Group-II vectors (at least one transition) (b) Group-III
vectors for zero CMV (at least two transitions) 66
Figure 4.8 MATLAB/Simulink modeling for implementation of improved
FCS-MPC using fuzzy normalization and space vector information for
switching decisions 71
Figure 4.9 MATLAB/Simulink modeling for implementation of predictive
delta sigma modulation 72
Figure 4.10 Experimental setup for validation of proposed modulation tech-
niques using inductive load 73
Figure 4.11 Simulated results for load voltage (vab ) and 3 − φ load currents
(iabc ) for (a) conventional FDM (b) Load voltage and 3 − φ load currents
for improvised FDM with SVM 74

xix
Figure 4.12 Experimental waveforms of Load voltage (vab ) and 3 − φ load
currents (iabc ) for (a) conventional FDM (b) Load voltage and 3 − φ load
currents for improvised FDM with SVM 75
Figure 4.13 Source voltage (vA ) , unfiltered source current (iA ) and filtered
source current (isA ); (a) conventional FDM (b) improvised FDM with SVM 76
Figure 4.14 Experimental waveforms of source voltage (vA ) , unfiltered source
current (iA ) and filtered source current (ia ) for (a) conventional FDM (b)
improvised FDM with SVM 76
Figure 4.15 Simulated harmonic spectrum of load current (ia ); (a) conven-
tional FDM (b) improvised FDM with SVM 77
Figure 4.16 Experimental harmonic spectrum of load current (isA ) in (a)
conventional FDM (b) improvised FDM with SVM 77
Figure 4.17 Simulated harmonic spectrum of filtered source current (isA ); (a)
conventional FDM (b) improvised FDM with SVM 78
Figure 4.18 Experimental harmonic spectrum of filtered source (isA ) in (a)
conventional FDM (b) improvised FDM with SVM 78
Figure 4.19 Simulated dynamic performance during load frequency change
for PDSM technique; (a) Load voltage and 3φ load currents (b) Source
voltage, reference and filtered source current. 81
Figure 4.20 Simulated dynamic performance during load amplitude change
for PDSM technique; (a) Load voltage and 3φ load currents (b) Source
voltage, reference and filtered source current. 81
Figure 4.21 Experimental waveforms of dynamic performance during load
frequency change for PDSM technique; (a) Load voltage and 3φ load cur-
rents (b) Source voltage, reference and filtered source current. 82
Figure 4.22 Experimental waveforms of dynamic performance during load
amplitude change for PDSM technique; (a) Load voltage and 3φ load cur-
rents (b) Source voltage, reference and filtered source current. 82

xx
Figure 4.23 Simulated steady state performance for PDSM technique; (a)
Load voltage and 3φ load currents (b) Source voltage, unfiltered and fil-
tered source current. 83
Figure 4.24 Experimental waveforms of system steady state performance for
PDSM technique; (a) Load voltage and 3φ load currents (b) Source voltage,
unfiltered and filtered source current. 83
Figure 4.25 Simulated harmonic performance for PDSM technique; (a) Har-
monic spectrum of load current (b) Harmonic spectrum of filtered source
current. 84
Figure 4.26 Experimental harmonic performance for PDSM technique; (a)
Harmonic spectrum of load current (b) Harmonic spectrum of filtered
source current. 84
Figure 4.27 Simulation based dynamic performance of the proposed algo-
rithm; (a) Change in frequency of reference load voltage (b) Change in
magnitude of reference load voltage 86
Figure 4.28 Experimental assessment of system dynamic performance; (a)
Change in frequency of reference load voltage (b) Change in magnitude of
reference load voltage 86
Figure 4.29 Simulated performance of delta sigma modulation for Group-III
vectors; (a) Load voltage control (b) Source current control 87
Figure 4.30 Steady state performance of DSM based MC for CMV elimina-
tion; (a) Load voltage and three-phase load currents (b) Source voltage,
unfiltered and filtered source current at near unity IPF 87
Figure 4.31 Experimental depiction of predictive delta sigma modulation for
CMV elimination; (a) Load voltage control (b) Source current control 88
Figure 4.32 Experimental waveforms is steady state of PDSM technique for
CMV elimination; (a) Load voltage and three-phase load currents (b)
Source voltage, unfiltered andf filtered source currents 88

xxi
Figure 4.33 Harmonic performance of DSM based MC; (a) Harmonic spec-
trum of load current (ia ) (b) Harmonic spectrum of filtered source current
(isA ) 89
Figure 4.34 Harmonic analysis (a) Harmonic spectrum of load current (b)
Harmonic spectrum of source current 89
Figure 4.35 Comparative analysis of source and load side current quality (a)
Space Vector Modulation (b) Mo-FCS-MPC (a) Proposed PDSM. 91
Figure 4.36 Output phase voltages and resulting common mode voltage; (a)
Using conventional 21 states (b) Using the DSM scheme with only Group-
III vectors 91

Figure 5.1 3φ-3φ (ABC-abc) matrix converter fed induction motor drive 96
Figure 5.2 Block diagrammatic representation for speed sensor-less vector
control of 3-φ IMD using conventional 21 states of matrix converter 106
Figure 5.3 Implementation of MO-FS-MPC with Fuzzy Normalized Errors 113
Figure 5.4 Block diagrammatic representation for speed sensor-less vector
control of 3-φ IMD using 6 CMV eliminating states of matrix converter 117
Figure 5.5 MATLAB/Simulink modeling for speed sensor-less vector control
of 3-φ IMD using conventional 21 states of matrix converter 118
Figure 5.6 MATLAB/Simulink modeling for speed sensor-less vector control
of 3-φ IMD using 6 CMV eliminating states of matrix converter 119
Figure 5.7 Simulated performance during starting 120
Figure 5.8 Simulated performance; (a) System dynamics while acquiring
near zero motor speed (b) System dynamics during speed reversal 121
Figure 5.9 Experimental performance; (a) Change in speed command (b)
Acquisition of near zero rotor speed 122
Figure 5.10 Experimental performance during speed reversal; (a) Torque and
speed response (b) Phase reversal of stator currents 122
Figure 5.11 Experimental performance during change in load torque 123

xxii
Figure 5.12 Simulated steady state performance of delta sigma modulator;
(a) Load voltage control (b) Source current control 123
Figure 5.13 Experimental steady state performance; (a) dq plot of stator flux
(b) Electromagnetic and load torque estimates (c) Reference and estimated
speed 124
Figure 5.14 Simulated harmonic performance; (a) Steady state stator current
(b) Harmonic spectrum of stator current (c) Steady state source current
at unity power factor (d) Harmonic spectrum of source current 125
Figure 5.15 Experimental harmonic performance; (a) 3-φ steady state stator
currents (b) Harmonic spectrum of stator current (c) Steady state source
current at unity power factor (d) Harmonic spectrum of source current 125
Figure 5.16 System dynamic performance during starting; (a) With precise
motor parameters (b) With 10% variation assumed in stator leakage in-
ductance 126
Figure 5.17 Experimental performance demonstrating motor control down
to zero rotor speed; (a) Electromagnetic torque (Te ), load torque estimate
(TL ) and rotor speed (Nr ) (b) 3-φ stator currents at different rotor speeds 127
Figure 5.18 Experimental performance during reversal in speed command;
(a) Rotor speed reversal (b) Dynamics in stator current during speed re-
versal 127
Figure 5.19 Simulated performance during load torque reversal 128
Figure 5.20 Experimental performance during change in load torque 129
Figure 5.21 Experimental steady state performance (a) dq plot of rotor flux;
(b) Estimated electro-magnetic torque (Te ) and load torque (TL ); (c) Pre-
dicted rotor speed (Nr ) and speed reference 130
Figure 5.22 Simulated steady state harmonic performance (a) stator current
profile and harmonic spectrum (b) source current profile and harmonic
spectrum 130

xxiii
Figure 5.23 Experimental steady state harmonic performance; (a) 3-φ stator
currents (b) Harmonic spectrum of stator current (c) Source current at
near unity IPF with respect to source voltage (d) Harmonic spectrum of
source current 131
Figure 5.24 CMV elimination using matrix converter; (a) Output phase volt-
ages with respect to star point of the source and resulting CMV (b) CMV
during device commutation 131
Figure 5.25 Comparative assessment of classical FS-MPC and FDM based
MO-FS-MPC; (a) Dynamic response during change in load torque for clas-
sical weight based MO-FS-MPC (b) Dynamic response during change in
load torque for FDM based MO-FS-MPC 133
Figure 5.26 Comparative assessment in terms of common mode voltage; (a)
CMV performance with classical 21 states (b) CMV performance with only
six states that trace a distinct input at each output. 134
Figure 5.27 Design of control gains k1 and k2 (a) ep.u
ω for different values of

k1 , and k2 = 0 (b) ep.u p.u

TL for different values of k2 , and k1 = 150 (c) eω for

simultaneous variation of k1 and k2 (d) ep.u

TL for simultaneous variation of

k1 and k2 . 135

Figure 6.1 3φ-3φ (ABC-abc) matrix converter fed induction motor drive 138
Figure 6.2 VSI output line to neutral voltage vectors 140
Figure 6.3 Output line to neutral voltage vectors in a matrix converter 141
Figure 6.4 Input current vectors and corresponding switching states in a
matrix converter 142
Figure 6.5 Block diagram of the proposed control algorithm 148
Figure 6.6 Block Diagram of open loop flux estimator and MRAC based
closed-loop speed observer 151
Figure 6.7 Block Diagram for Direct Torque Control of Induction Motor
using Space Vector Modulated Matrix Converter 153

xxiv
Figure 6.8 Space vector diagrams corresponding to output voltage (L-L)
vectors and input current vectors 154
Figure 6.9 MATLAB/Simulink modeling for speed sensor-less hysteresis
based DTC of 3-φ IMD using matrix converter using an adaptive observer 157
Figure 6.10 MATLAB/Simulink modeling for speed sensor-less DTC-SVM
of 3-φ IMD using matrix converter using MRAS observer 157
Figure 6.11 Various estimated parameters during starting 159
Figure 6.12 Various system parameters during starting 160
Figure 6.13 System parameters during reversal of speed command 161
Figure 6.14 Experimental demonstration of speed reversal 161
Figure 6.15 System parameters during load perturbation 162
Figure 6.16 System dynamics during source perturbation 162
Figure 6.17 DQ variables for stator fluxes 163
Figure 6.18 Experimental DQ plot of stator flux 163
Figure 6.19 Experimental steady state waveforms for hysteresis based DTC;
(a) Stator voltage and three-phase stator currents (b) Grid voltage and
grid current at near unity power factor 164
Figure 6.20 Experimental steady state waveforms; (a) Hysteresis control of
torque (b) Reference and estimated rotor speed 164
Figure 6.21 Unfiltered source current and its harmonic spectrum 164
Figure 6.22 Filtered source current and its harmonic spectrum 165
Figure 6.23 Stator Flux, torque and speed estimation from stator voltage
and sensed stator current 166
Figure 6.24 Dynamic behaviour of system parameters during starting 166
Figure 6.25 System Dynamics during speed reversal 167
Figure 6.26 Experimental steady state waveforms for DTC-SVM; (a) Stator
voltage and three-phase stator currents (b) Grid voltage and grid current
at near unity power factor 168

xxv
Figure 6.27 Experimental steady state speed response and stator flux angle
estimation 168
Figure 6.28 System Dynamics at varying speeds for (a) Conventional DTC;
(b) DTC SVM 169
Figure 6.29 Unfiltered Source Current and its harmonic spectrum for (a)
Conventional DTC; (b) DTC SVM 170

Figure 7.1 1φ-3φ matrix converter fed induction motor drive 173
Figure 7.2 Delta-Sigma Modulation for a 1φ-3φ matrix converter 175
Figure 7.3 System block diagram for low speed sensor-less load oriented
control of single-phase to three-phase matrix converter fed induction motor
drive 178
Figure 7.4 MATLAB/Simulink modeling for load side power quality im-
provement in 1-φ to 3-φ MC fed RL load, using delta sigma modulation. 179
Figure 7.5 MATLAB/Simulink modeling for speed sensor-less control of 1-φ
to 3-φ MC fed IMD using MC with improved motor current quality. 180
Figure 7.6 Simulated dynamic performance of delta-sigma modulated 1φ −
3φ MC during load frequency change 182
Figure 7.7 Experimental dynamic performance of delta-sigma modulated
1φ − 3φ MC during load frequency change 182
Figure 7.8 Delta-Sigma Modulation for a 1φ-3φ matrix converter 183
Figure 7.9 Simulated waveforms of source voltage, unfiltered and filtered
source currents 183
Figure 7.10 Experimental waveforms of source voltage, unfiltered and filtered
source currents 184
Figure 7.11 Simulated harmonic spectrum of load current for DSM modu-
lated 1φ − 3φ MC. 184
Figure 7.12 Experimental harmonic spectrum of load current for DSM mod-
ulated 1φ − 3φ MC. 184
Figure 7.13 Simulated harmonic spectrum of filtered source current 185

xxvi
Figure 7.14 Experimental harmonic spectrum of filtered source current 185
Figure 7.15 Simulated starting performance of DSM modulated 1φ-3φ MC
fed IMD 186
Figure 7.16 Simulated system dynamics during speed control (100 rpm to
near 0 rpm) in a DSM modulated 1φ-3φ MC fed IMD 187
Figure 7.17 Experimental system dynamics during speed control (100 rpm
to 45 rpm to 15 rpm) in a DSM modulated 1φ-3φ MC fed IMD 188
Figure 7.18 Simulated system dynamics during speed reversal in a DSM mod-
ulated 1φ-3φ MC fed IMD 188
Figure 7.19 Experimental system dynamics during speed reversal in a DSM
modulated 1φ-3φ MC fed IMD (a) Speed reversal (b) Phase reversal in
stator currents 189
Figure 7.20 Simulated system dynamics during sudden withdrawal of load
in DSM modulated 1φ-3φ MC fed IMD 189
Figure 7.21 Experimental system dynamics during increase and subsequent
decrease of load in DSM modulated 1φ-3φ MC fed IMD 190
Figure 7.22 Simulated DQ plot of stator flux 191
Figure 7.23 Experimental DQ plot of stator flux 191
Figure 7.24 Experimental steady state performance of the drive (a) PWM
load voltage and 3-φ motor currents (b) Reference and predicted motor
speed (c) Load torque and predicted electro-magnetic torque. 191
Figure 7.25 Stator current and its harmonic profile 192
Figure 7.26 Source current and its harmonic profile 192
Figure 7.27 Grid side harmonic performance (a) Source voltage and source
current (b) Harmonic spectrum of 1-φ grid current 193

Figure 8.1 Desired waveforms in ideal 1-φ to 3-φ power conversion; (a) 1-φ
source side parameters (b) 3-φ load side parameters 195
Figure 8.2 1φ-3φ matrix converter fed induction motor drive 196

xxvii
Figure 8.3 Performance of Kahn’s modulation technique; (a) System pa-
rameters with resistive load (b) System parameters with inductive load 197
Figure 8.4 MATLAB/Simulink modeling for implementation of Mo-FCS-
MPC in 1-φ to 3-φ MC fed inductive load. 203
Figure 8.5 MATLAB/Simulink modeling for speed sensor-less Mo-FCS-MPC
of 1-φ to 3-φ MC fed IMD. 204
Figure 8.6 Simulated dynamic performance during change in reference load
current (or load voltage) amplitude; (a) SPWM (b) Load Oriented FCS-
MPC (c) MO-FCS-MPC. 206
Figure 8.7 Simulated dynamic performance during change in reference load
frequency from 5Hz to 10Hz; (a) SPWM (b) Load Oriented FCS-MPC (c)
MO-FCS-MPC. 206
Figure 8.8 Experimental dynamic performance during change in reference
load current (or load voltage) amplitude; (a) SPWM (b) Load Oriented
FCS-MPC (c) MO-FCS-MPC. 207
Figure 8.9 Experimental dynamic performance during change in reference
load current frequency; (a) SPWM (b) Load Oriented FCS-MPC (c) MO-
FCS-MPC. 207
Figure 8.10 Simulated steady state performance; (a) SPWM (b) Load Ori-
ented FCS-MPC (c) MO-FCS-MPC. 208
Figure 8.11 Experimental steady state performance on the load side; (a)
SPWM (b) Load Oriented FCS-MPC (c) MO-FCS-MPC. 208
Figure 8.12 Experimental steady state performance on the source side; (a)
SPWM (b) Load Oriented FCS-MPC (c) MO-FCS-MPC. 208
Figure 8.13 Simulated load side harmonic performance; (a) SPWM (b) Load
Oriented FCS-MPC (c) MO-FCS-MPC. 209
Figure 8.14 Simulated source side harmonic performance; (a) SPWM (b)
Load Oriented FCS-MPC (c) MO-FCS-MPC. 209

xxviii
Figure 8.15 Experimental source side harmonic performance; (a) SPWM (b)
Load Oriented FCS-MPC (c) MO-FCS-MPC. 210
Figure 8.16 Source current displacement with respect to source voltage; (a)
SPWM (b) Load Oriented FCS-MPC (c) MO-FCS-MPC. 210
Figure 8.17 Starting performance of MO-FCS-MPC based 1-φ to 3-φ MC fed
IMD (a) Starting with MO-FCS-MPC (b) Starting with single objective
load oriented FCS-MPC, followed by switching over to MO-FCS-MPC. 212
Figure 8.18 Dynamic performance of MO-FCS-MPC based 1-φ to 3-φ MC
fed IMD 213
Figure 8.19 Experimental performance of MO-FCS-MPC based 1-φ to 3-φ
MC fed IMD 213
Figure 8.20 Steady state performance of MO-FCS-MPC based 1-φ to 3-φ
MC fed IMD 214
Figure 8.21 Steady state performance of MO-FCS-MPC based 1-φ to 3-φ
MC fed IMD (a) Torque response (b) Speed Response (c) DQ plot of rotor
flux 214
Figure 8.22 Motor side power quality assessment in MO-FCS-MPC based
1-φ to 3-φ MC fed IMD 215
Figure 8.23 Source side power quality assessment in MO-FCS-MPC based
1-φ to 3-φ MC fed IMD (a) Source current and its harmonic spectrum (b)
Operation near unity input power factor. 216
Figure 8.24 Experimental waveform on the motor side 216
Figure 8.25 Source side power quality assessment in MO-FCS-MPC based 1-
φ to 3-φ MC fed IMD (a) Source current at near unity input power factor.
(b) Harmonic spectrum of source current 216

xxix
List of Tables

Table 3.1 Induction Motor Parameters 36

Table 4.1 Switching States and resulting CMV in Matrix Converter 48

Table 4.2 Example of Switching Table Generation from Sector Information 55
Table 4.3 Example of Load Voltage Modulator (LVM) Implementation using
DSM for Group-I and Group-II vectors 59
Table 4.4 Example of Source Current Modulator (SCM) Implementation using
DSM for Group-I and Group-II vectors 61
Table 4.5 LVM Switching Implementation for Load Voltage Control for Group-
III vectors 68
Table 4.6 SCM Switching Implementation for Source Current Control for Group-
III vectors 70
Table 4.7 System Parameters and Their Specifications 74
Table 4.8 Choice of Control Parameter ‘n ’ 79
Table 4.9 System Parameters used in Simulation/Experimentation 80
Table 4.10 System Specifications 85
Table 4.11 Comparison of proposed PDSM technique with other modulations
schemes for MCs 90
Table 4.12 Comparative performance of existing modulation techniques for Group-
III vectors 92

Table 5.1 Load Voltage Control (LVC) through DSM 105

Table 5.2 Source Current Control (SCC) Through DSM 106
Table 5.3 Zero CMV states in Matrix Converter 107
Table 5.4 Comparison between harmonic performance of classical FS-MPC
and FDM based FS-MPC at different loading conditions 133

Table 6.1 Switching Table For DTC 141

xxx
Table 6.2 Switching Combinations and Resultant Space Vectors in a MC 143
Table 6.3 Switching Table for DTC via MC incorporating PFC 144
Table 6.4 Switching Combinations and Resultant Space Vectors in a MC 155
Table 6.5 Example of Switching Table Generation for SVM with IPF control 156
Table 6.6 Source Current Parameters at varying motor speeds in conventional
DTC and DTC SVM 170

Table 7.1 Switching Combinations for a 1φ − 3φ MC 176

Table 7.2 Example of Decisional Element Implementation in Delta-Sigma Mod-
ulated 1φ − 3φ MC 177
Table 7.3 System Specifications For Delta Sigma Modulation in 1φ-3φ MC 181

Table 8.1 Switching States in 1t3-MC 200

Table 8.2 Specifications of System Parameters 205

xxxi